The present invention relates generally to a system and method for providing connectivity between networks, and more particularly, to a system and method for providing connectivity with collision detection in large-scale networks.
Data networks, in general, use multiple layers of communication protocols to effectuate data communication between entities on the network. The lowest layer of the communication is often referred to as the physical layer. The second layer is often referred to as the packet layer. Communication standards that include such multiple layer connectivity include that defined in the ISO 8802/3IEEE 802.3 specification for 10Base-T local area networks.
In accordance with such communication schemes, lower layers are generally employed for local switching between the network entities connected to a single hub. In general, physical layer switches are geographically limited in part because of the methodologies employed to detect whether connectivity is available. According to the ISO standards, a source entity determines physical layer connectivity by sending a packet to the hub, the packet being intended for a destination entity. When the hub receives a transmit packet, it repeats the packet to all entities that are connected to the hub. If another network entity has transmitted a packet to the hub before the packet from the source hub is completely received by the hub, the source entity detects a collision and then determines that the transmission is unsuccessful. If however, no collision is detected, the hub provides the connection to the destination entity and passes the transmitted packet directly through.
Packet layer switching, which typically occurs between hubs of a larger network, includes the step of sending one or more packets to a packet switch from a source entity. The packet switch then stores one or more packets and transmits the packets when connectivity to the destination entity or another intermediate switch is available. By contrast, in physical layer switching, as discussed above, the collision is made in real-time as the source entity packet is being transmitted.
Accordingly, physical layer switching allows for faster communication than packet layer switching because physical layer switching does not involve the storage of packets in the intermediate switch. However, packet layer switching is usually employed to establish connectivity between multiple local area networks (xe2x80x9cLANsxe2x80x9d). Thus, communication between entities on multiple local area networks is relatively slow as compared to communication between entities on the same local area network.
A switching system has been proposed, however, that allows multiple LANs to be connected at physical layer, thus providing increased communication speed. The switching system is described in U.S. patent application Ser. No. 09/203,016, filed Nov. 30, 1998, now U.S. Pat. No. 6,353,858 issued March 5, 2002. which is assigned to the assignee of the present invention and incorporated herein by reference. The system includes a space switching unit and a plurality of switch interface units coupled between the space switching unit and a plurality of LANs. When a LAN provides a transmit packet to its switch interface unit, the switch interface unit establishes a first unilateral path from the destination entity to the space interface unit that is coupled to the source entity. If the space interface unit detects activity on the first unilateral path, the space interface unit provides a collision indication to the source entity before the source entity has finished transmitting the transmit packet. Because the collision is provided before the source has finished transmitting the packet, the source entity logs a collision as it would in any LAN collision.
If, however, the switch interface unit detects no activity on the first unilateral path, the switch interface unit establishes a second unilateral path from the source entity to the destination entity to allow communications. A first-in-first-out buffer or the like delays the transmit packet a sufficient amount of time to allow the collision determination to be made.
Thus, the entire connection operation described in the U.S. patent application Ser. No. 09/203,016 now U.S. Pat. No. 6,353,858 issued Mar. 5. 2002. is provided within the standard communication requirements of a physical layer switching operation. As a result, connectivity between multiple entities on multiple LANs may be accomplished relatively quickly.
While the forgoing switching system can increase transmission speed between LANs, it is limited by the practical number of connections that the space switching unit may make. The space switching unit typically is an integrated circuit that allows each of m inputs to be connected to each of m outputs. Currently, such a device allows for first and second unilateral connections (i.e. transmit and receive links) between 128 entities. Each of the links is independently addressed through corresponding m single input address lines. Although such a device may be expanded to provide 256 or more connections, the number of connections remains limited to the capacity of the space switching unit.
Consequently, there is a potential need for expand physical layer switching capacities in a switching system between multiple LANs (or other sub-networks).
The present invention fulfills the above need(s), as well as others, by providing a switching arrangement that allows for physical layer switching between large numbers of Hubs. To this end, the switching arrangement includes a cross point matrix having a plurality of column control busses that interconnect the control circuitry of one or more columns of cross points. The column control busses eliminate the need for individual address control connections from each source port to each possible destination port cross point. As a result, large numbers of cross points are possible. Moreover, because each port need not be connected to the control circuitry each cross point, multiple cross point integrated circuits may be combined to increase switching capacity.
An exemplary embodiment of the present invention includes a method for communicating from a source port (i) to a destination port (j). The method is used with a switching system that has m ports, each of the ports being coupled to a local area network via a Hub. The connectivity between the inputs and outputs of the m ports forms a matrix of cross points having m rows and m columns. Each port has a transmit line being coupled to a row of the matrix and a receive line being coupled to a column of the matrix. A transmission operation from the source port (i) to the destination port (j) involves a first control circuit for unilaterally connecting the port (i) to the port (j) and a second control circuit for unilaterally connecting the port (j) to the port (i) (where i or j=1, 2, . . . , m).
The method comprises a first step of sending address information from the port (i) to a third control circuit. The method also includes the step of providing a column control bus that couples the third control circuit to a plurality of control circuits including the second control circuit, each of the plurality of control circuits operable to control the operation of a cross point. Then the address information is routed from the third control circuit to all of the plurality of control circuits including the second control circuit through the column control bus. The method further includes the step of making a unilateral path connection at a return path cross point from the destination port (j) to the source port (i) responsive to receiving the address information at the second control circuit.
The return path cross point allows monitoring the destination hub to determine if it is idle. Such monitoring is necessary to carry out physical layer switching. In particular, if the hub is not idle, a message may be sent to the source hub before the source hub completes transmission of the packet.
The present invention also provides a corresponding apparatus to perform the steps in the above-described method.